Workload Allocation in Hierarchical Cloudlet Networks

Edge cloudlets are promising to mitigate the high network delay incurred by the remote cloud in executing workloads offloaded from a user equipment (UE). However, the response time of a task request consists of both the network delay and computing delay. Considering the spatial and temporal dynamics of workloads among cloudlets, if the workload of an edge cloudlet is heavy, the computing delay in the cloudlet may be unbearable. In this letter, we design a hierarchical cloudlet network and propose a workload allocation scheme to minimize the average response time of UEs' requests by deciding which cloudlet a UE is assigned to and how much computing resource is provisioned to serve it. The performance of the proposed scheme is validated by extensive simulations.